The following figure illustrates a possible output scenario when using more than a single band. In a multiband system, it is possible to have different number of antennas on each band. In this example, RCID 4 and RCID 7 are extracting PRACH data from an eight-antenna band, and RCID 9 is extracting PRACH data from a band with four antennas.
The first samples from RCID 7 are not coincident with any other RCID output. So, they appear as a contiguous block of eight samples across two lanes. The second sample from RCID 7 is coincident with the other RCID outputs. This results in the samples being interleaved with the other RCIDs. In this scenario, the TID sequence for the RCID 7 samples is preserved. However, the spacing between samples cannot be predicted.
RCID 4 does not require the sample order to be the same as RCID 7. The antennas come out in an order dictated by internal arbitration.
RCID 9 has half the number of output samples because it comes from a band with four active antennas. The four antennas are allocated across two output lanes, hence there are two TID values. Due to internal constraints, even though there are only four antennas for this RCID, the channel outputs as if there are eight. It inserts a "null" sample in the place of the unused antennas. This results in a cycle with TVALID Low where there would have been a sample had the RCID been acting on eight antennas.
The band with the most antennas when the core instance is generated determines how many samples are generated per RCID. If an RCID is capturing from a band with fewer antennas, there are null samples in the output. This applies even if no RCID is capturing PRACH data from the band with the largest number of antennas.
If all bands have the same number of antennas, there are no null samples. In this scenario, the output is indistinguishable from a single band case.